JPH07103751A - Method for measuring roll profile - Google Patents

Abstract

PURPOSE:To obtain a highly accurate profile by moving a plurality of distance sensors via equal intervals by the interval, obtaining a roll profile for every discrete point when tone sensors are stopped before and after moving the distance, and interpolating the interval by distance data obtained during the movement. CONSTITUTION:A sensor stage 3 is moved left along a moving guide 4 and stopped. Many distance values are measured for a predetermined time by distance sensors 2a-2e of (n) sensors set on the stage 3 via equal intervals (d). Far different values due to an accidental error are removed from the measured values, thereby to obtain an average value Xa(i) for every sensor 2a-2e (i=1-n, and (n) is the number of sensors.). Then, the stage 3 is moved right by the distance (d) and stopped and, an average value Xb(i) of the measured distances for a predetermined time is similarly obtained for every sensor 2a-2e. Moreover, measured distance values Xj(i) of the sensors 2a-2e and corresponding moving distances D(j) are obtained during the movement to interpolate a schematic shape of a roll profile based on the average values Xa, Xb by the measured values Xj, D (j=1-m, (m) is the number of data.).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for measuring the profile of a roll such as a steel rolling roll.

[0002]

2. Description of the Related Art The surface profile distribution in the axial direction of a rolling roll used in the rolling process of steel and the like, that is, the roll profile, is constantly changing due to thermal expansion of the rolling material due to processing heat or deformation due to friction with the rolling material. And changes. Since this roll profile has a great influence on the shape of the material to be rolled, it is possible to measure this change online or in-line and control the roll bending force and roll cooling water amount to appropriate values based on the measured values. is important.

As a method for measuring the roll profile,
Generally, the sensor mount is placed almost parallel to the roll axis, and while moving this mount parallel to the roll axis,
A method is known in which the distance distribution to the roll surface is measured by a plurality of distance sensors mounted on a sensor mounting base, and the roll profile is obtained from the output of each sensor. However, in reality, the displacement of the sensor mount (hereinafter referred to as "sensor posture change") at each measurement timing is unavoidable due to deformation of the sensor mount movement guide, backlash when the sensor mount moves, etc. Therefore, it is necessary to correct this and measure the roll profile. The method of measuring the roll profile by correcting the sensor attitude change is
For example, Japanese Patent Application Laid-Open No. 61-159106 and Japanese Patent Application Laid-Open No. 3-1
It is proposed in Japanese Patent No. 86712.

The method proposed in Japanese Unexamined Patent Publication No. 61-159106 uses four or more distance sensors arranged on the sensor mount at equal intervals d, and the distance is changed every time the sensor mount moves a distance d. After obtaining the measured values, the sensor attitude change at each measurement timing is expressed by using two variables, and simultaneous equations representing the roll profile, the total distance measured values and the relationship between these variables are made, and this is solved by the least squares method. Is the method.

In the method proposed in Japanese Patent Laid-Open No. 3-186712, a plurality of distance sensors and a set of three correction sensors for measuring sensor attitude changes are mounted on a sensor mounting base to mount the sensors. This is a method of obtaining a roll profile by obtaining a distance measurement value when the table is moved and correcting the distance sensor measurement value. This method uses the "three-point method" that is used for roundness measurement, etc. to calculate the sensor posture change by the correction sensor. However, in the "three-point method", the quadratic function component (parabola) of the roll profile is used. The amount of deformation is not calculated in principle, so the roll profile is known before the measurement.
It is devised so that a roll having a quadratic function component can be measured and a quadratic function component can be obtained.

[0006]

However, the above-mentioned conventional method has the following problems. That is, the distance measurement value is obtained while moving the sensor mount, and the measurement by the one distance sensor at one position in the roll axis direction can be performed only once. In this case, if some accidental measurement error occurs in a certain distance measurement value, it cannot be compensated. Particularly in the online measurement, it is necessary to measure the rotating roll, and eccentricity of the roll and play due to roll rotation also cause such an error in distance measurement.

Further, when the roll profile is calculated by correcting the sensor attitude change, the influence of the above distance measurement error is not limited to appearing at only one position where the error occurs,
There was a problem that it extends to the entire calculated roll profile. In view of the above circumstances, it is an object of the present invention to provide a roll profile measuring method capable of obtaining a roll profile with high accuracy.

[0008]

In order to achieve the above-mentioned object, the present invention mounts a plurality of distance sensors on a sensor mounting base arranged substantially parallel to the roll axis, and the sensor mounting base is substantially aligned with the roll shaft. In the roll profile measuring method, the distance distribution to the roll surface is measured by each distance sensor while moving in parallel, and the axial profile of the roll is obtained based on the output of each distance sensor.
The distance sensor of 3) is mounted on the sensor mounting base at regular intervals d such that n · d ≧ L with respect to the roll profile measurement length L, and (2) the sensor mounting base is mounted at one end in the roll axial direction. In the stationary state, the distance measurement value is obtained by each distance sensor for a predetermined time, and the distance data Xa (i) (i = 1) at each position of each of n distance sensors is obtained based on these distance measurement values. , 2, ..., N), and (3) while moving the sensor mount toward the other end in the roll axis direction, the distance measurement value Xj (i) (i = 1,
2, ..., n; j = 1,2, ..., m), and (4) the sensor mount is moved by the distance d and then stopped, and a distance measurement value is obtained by each distance sensor for a predetermined time. Distance data Xb (i) (i = 1, 2, ..., N) at each position of each of n distance sensors based on the distance measurement value of
(5) Show a rough shape of the roll profile based on the distance data Xa (i) and the distance data Xb (i) (n +
1) The roll profile value of each of the discrete points is obtained, and (6) the profile is obtained by interpolating between the roll profile values of the discrete points based on the distance data Xj (i). This is a roll profile measuring method.

[0009]

According to the roll profile measuring method of the present invention, the distance is measured by a predetermined time before and after the movement of the sensor mount, so that the distance d is increased by (n + 1).
For each discrete point, the profile value for that roll is accurately measured. Further, for each intermediate point between the discrete points, the roll profile values at the discrete points are interpolated based on the distance data Xj (i) measured while moving the sensor mount, so that as a whole. Highly accurate roll profile is required.

[0010]

EXAMPLES Examples of the present invention will be described below. FIG. 1 is used for the roll profile measuring method of the present invention.
It is the figure which showed the sensor mounting stand in which several distance sensors were arrange | positioned, and the roll of to-be-measured object. The number n of distance sensors and the arrangement interval d of these distance sensors can be arbitrarily determined within a range of n ≧ 3 and n · d ≧ L, where L is the measurement roll length. FIG. 1 shows an example in which n = 5. (1) First, the sensor mount 3 is moved to one side (the left end in FIG. 1) along the movement guide 4 and stopped. (2) With each of the distance sensors 2a, 2b, 2c, 2d, 2e,
Many distance measurement values are obtained during a predetermined time t ₀ , and the average value Xa (i) (i = 1, 2, ..., N) is obtained for each distance sensor.
Ask for. At this time, if there is a value that is far from other distance measurement values, it is regarded as a value caused by some random error, and that value is excluded, and the average value is calculated. Measurement time t ₀
Is 1 when the natural frequency of the sensor mount 3 is ν
If it is decided to be an integral multiple of / ν, the sensor mount 3
This is preferable since the measurement error due to the vibration can be significantly reduced. Further, in the case of measurement in a roll rotating state, if the measurement time t ₀ is set to correspond to a time that is an integral multiple of the roll rotation period, the influence of the eccentricity of the roll 1 or the influence of backlash synchronized with the roll rotation will be remarkable. There is an effect that can be reduced. In this way, very reliable distance data Xa (i) that is not affected by the eccentricity of the roll 1 or the vibration of the sensor mounting base 3 as well as the random error of the distance sensors 2a to 2d can be obtained. (3) Next, while moving the sensor mount 3 to the right in FIG. 1, the distance measurement values Xj are measured by the distance sensors 2a to 2e.
(I) (i = 1, 2, ..., N; j = 1, 2, ..., M), and at the same time, the moving distance data D (j) (j = 1, 1) of the sensor mount 3 at each measurement timing. 2, ..., m)
Is obtained (0 ≦ D (j)). Where j is the sensor mount 3
Is a number representing the measurement order of each of the distance sensors 2a to 2e obtained while moving, and m is the number of data obtained by each of the distance sensors 2a to 2e while the sensor mount 3 moves by the distance d. Is. (4) When the sensor mount 3 moves by the distance d, the sensor mount 3 is stopped. (5) While the sensor mount 3 is stopped, each distance sensor 2a to 2e obtains many distance measurement values for a predetermined time t ₀ , and similarly to (2) above, the averaged distance is obtained. Data Xb (i) (i = 1, 2, ..., N) is obtained. (6) Distance data Xa obtained in (2) and (5) above
Of (i) and Xb (i), Xa (k + 1) and Xb
Since (k) (k = 1, 2, ..., N-1) is the distance measurement value for the same position of the roll, it should match if there is no error. As shown in 2, the values are slightly different. Therefore, follow the steps (a) to (c) below to correct the change in sensor attitude,
Roll profile values Y (i) of (n + 1) discrete points
(I = 1, 2, ..., N + 1) is obtained (FIG. 2 shows an example in the case of n = 8). (A) First, conversion is performed to remove the difference in sensor attitude change before and after the sensor mount is moved. Here, it is assumed that the change in posture of the sensor mount after the movement of the sensor mount is made to coincide with that before the movement:

[0011]

[Equation 1]

Α and β are obtained by the least-squares method so as to minimize. (B) Corrected distance data Xb ′ obtained in this way
(I) and distance data Xa (i) before moving the sensor mount
Z (1) = Xa (1) Z (i) = {Xa (i) + Xb ′ (i−1)} / 2 where i = 2, 3, ..., N Z (n + 1) = Xb '(N) gives distance data Z (i) = (n + 1) discrete points
(I = 1, 2, ..., N + 1) is calculated. (C) Next, the change in the attitude of the sensor mount before the movement is corrected. Here, since it is sufficient for the roll profile to know the uneven distribution on the roll surface, the amount of displacement from an arbitrary reference line may be expressed. Therefore, if we select a reference line that makes the points at both ends in the roll axis direction zero,
You can do the following conversions:

[0013]

[Equation 2]

The roll profile values Y (i) of the (n + 1) discrete points thus obtained (i = 1, 2, ...,
n + 1) is highly reliable distance data Xa (i), Xb
Since it is obtained from (i), the distance sensors 2a-2
It is a value that is not affected by the random error included in each measured value of e, the eccentricity of the roll 1, the vibration of the sensor mounting base 3, and the like, and the sensor attitude change is also corrected by the above calculation. Therefore, Y (i) is a highly accurate measurement value that accurately represents the (n + 1) discrete points of the roll profile.

If the number n of distance sensors is increased and the sensor arrangement interval d is narrowed, a large number of such highly accurate discrete points can be obtained. On the other hand, if the number of sensors is increased, it becomes economically expensive. (7) Finally, between the roll profile values Y (i) of the (n + 1) discrete points, the measured value X while the sensor mount is moving.
j (i) (i = 1, 2, ..., N; j = 1, 2, ..., M)
To obtain a continuous roll profile in the roll axis direction. An example of this interpolation method is the following procedure (a)-
It shows in (d). (A) Assuming that the values caused by the change in sensor attitude at each measurement timing j (j = 1, 2, ..., M) are u and v, the data Xj ′ (i) after the change in sensor attitude is expressed by the following equation. To be done.

Xj ′ = Xj (i) + u · i + v (1) The values u and v for each j are obtained by the procedure from (b) below. (B) Roll profile value Y of (n + 1) discrete points
(I) (i = 1, 2, ..., N + 1) is an M-th order (M ≧ 2) polynomial

[0017]

[Equation 3]

Is approximated by The calculation can be easily performed by the method of least squares. (C) For each j,

[0019]

[Equation 4]

The values u and v that minimize are obtained by the method of least squares. (D) From the obtained values u and v, Xj ′ is obtained from the above equation (1).
(I) is determined, and by connecting the data groups Y (i) and Xj ′ (i), a roll profile in which the change in sensor attitude is corrected can be obtained. Note that the interpolation method of the roll profile value Y (i) at discrete points by the measured value Xj (i) is not limited to the above method, and various conventionally proposed methods such as the above-mentioned "three-point method" can be used. The method using can be used. In this case, the outline shape of the roll profile has already been obtained as Y (j),
It is not necessary to measure the calibration roll in advance and obtain the quadratic function component of the roll as in the conventional method described above.

According to the above procedures (1) to (7), the roll is made by the roll profile value of (n + 1) discrete points which is not affected by the random error of the distance sensor, the eccentricity of the roll, the vibration of the sensor mount or the like. Since the outline of the profile is obtained with high accuracy and the gap is interpolated to obtain the roll profile, the entire roll profile is not distorted by the measurement error of the distance sensor,
An accurate roll profile is obtained.

An example of simulation using the roll profile measuring method according to the present invention is shown in FIG. In this example, the profile of a roll having a roll crown of 50 μm was used for the method of the present invention and the conventional method (method utilizing the three-point method).
In this example, the data of one distance sensor in the middle of measurement is intentionally set to 5 μm (imagining a random error).
It is changing. In the method according to the present invention, the influence of this random error is slight, but in the conventional method, the entire roll profile is distorted.

[0023]

According to the present invention, when a plurality of distance sensors are moved to measure a roll profile, the distance sensors are stopped before and after the movement to obtain a large amount of distance data at a predetermined time, resulting in an accidental error. Since a highly accurate discrete roll profile that is not affected by is first obtained and the gap is interpolated by the moving distance data, the entire roll profile is not distorted by accidental measurement error of the distance sensor, and It is possible to reduce the influence of eccentricity and play when the roll rotates, and a highly accurate roll profile is required.

[Brief description of drawings]

FIG. 1 is a schematic diagram for explaining an example of a roll profile measuring method of the present invention.

FIG. 2 is a schematic diagram showing measured distance data.

FIG. 3 is a diagram showing a simulation example of the present invention.

[Explanation of symbols]

 1 Rolls 2a to 2e Distance sensor 3 Sensor mount 4 Sensor mount movement guide

Front page continued (72) Inventor Fumihiko Ichikawa 1 Kawasaki-cho, Chuo-ku, Chiba Kawasaki Steel Co., Ltd. Technical Research Headquarters (72) Inventor Tomoaki Kimura 3-1-1 Sachimachi, Hitachi City, Ibaraki Hitachi Hitachi, Ltd. Heavy Equipment Design Department

Claims (1)

[Claims] 1. A plurality of distance sensors are mounted on a sensor mount arranged substantially parallel to the axis of the roll, and each distance sensor moves the sensor mount to the roll surface while moving the sensor mount along the axial direction of the roll. In a roll profile measuring method of measuring a distance distribution and obtaining a profile in the axial direction of a roll based on the output of each distance sensor, n (n ≧ 3) distance sensors are used for the roll profile measuring length L, and n · d. Mounted on the sensor mount at equal distances d such that ≧ L, and while the sensor mount was moved to one end in the roll axis direction, the distance measurement value was obtained with each distance sensor for a predetermined time. , Distance data Xa (i) (i = 1, 2, ..., At each position of each of the n distance sensors based on these distance measurement values.
n), and while moving the sensor mount toward the other end in the roll axis direction, the distance measurement value Xj (i) (i =
1, 2, ..., N; j = 1, 2, ..., M), the sensor mount is moved by a distance d and then stopped, and a distance measurement value is obtained by each distance sensor for a predetermined time. Distance data Xb (i) (i = 1, 2, ..., N) at each position of each of the n distance sensors is obtained based on the distance measurement value, and the distance data Xa (i) and the distance data Xb (i) are obtained. Shows the general shape of the roll profile based on (n + 1)
A roll profile measuring method, wherein a roll profile value of each of the discrete points is obtained, and the profile is obtained by interpolating between the roll profile values of the discrete points based on the distance data Xj (i).